← mapFunctions & Growth

Exponential Functions

⚗ Dr. Möbius, from the lab

You have been dealing with polynomials, where every step adds a fixed amount. Now I'm introducing the thing that eats polynomials for breakfast, stacks the bones in the corner, and doesn't even look up. An exponential function multiplies by the same ratio every single step. One sentence. That's the whole goddamn concept. The consequences, as you are about to discover, are frankly obscene — and I use that word with professional admiration.

THE BIG IDEA

An exponential function grows (or decays) by a constant multiplicative ratio at every step, which is fundamentally different from the constant additive step of linear growth — and that difference eventually dominates everything.

Constant ratio: the defining property

Here is the distinction that matters.

A linear function f(x)=mx+bf(x) = mx + b has a constant difference between outputs: f(x+1)f(x)=mf(x+1) - f(x) = m, regardless of where you are on the xx-axis.

An exponential function f(x)=abxf(x) = a \cdot b^x has a constant ratio between outputs: f(x+1)f(x)=abx+1abx=b.\frac{f(x+1)}{f(x)} = \frac{a \cdot b^{x+1}}{a \cdot b^x} = b.

Every time xx increases by 11, you multiply the output by bb. That's it. bb is called the base, and it's the whole engine.

This matters because multiplication compounds and it does not stop. After nn steps, starting from aa, you have abna \cdot b^n. After nn steps of linear growth from aa with rate mm, you have a+mna + mn. The linear function adds mnmn total; the exponential multiplies bnb^n times. For b>1b > 1 and large nn, these are in completely different universes — one of them is still in the parking lot while the other is orbiting Jupiter.

The graph of bxb^x

Let f(x)=bxf(x) = b^x with b>0b > 0, b1b \ne 1.

Growth: b>1b > 1.

  • Always positive: bx>0b^x > 0 for all xx.
  • Passes through (0,1)(0, 1) always: b0=1b^0 = 1.
  • As x+x \to +\infty: f(x)+f(x) \to +\infty (steep growth).
  • As xx \to -\infty: f(x)0+f(x) \to 0^+ (approaches the xx-axis from above, never touching it).
  • Horizontal asymptote: y=0y = 0.

Decay: 0<b<10 < b < 1.

  • Same shape, reflected horizontally. Now: as x+x \to +\infty, f(x)0+f(x) \to 0^+; as xx \to -\infty, f(x)+f(x) \to +\infty.
  • Same horizontal asymptote y=0y = 0.

Drag the grapher to see growth (b=2b = 2) and decay (b=1/2b = 1/2) side by side:

function grapher
x = 1.22^x → 2.2970.5^x → 0.435

Compound interest: the canonical story

Suppose you deposit principal PP in an account paying annual interest rate rr (as a decimal), compounded nn times per year. After tt years, the balance is: A=P ⁣(1+rn)nt.A = P\!\left(1 + \frac{r}{n}\right)^{nt}.

Why? After one compounding period, you have P(1+r/n)P \cdot (1 + r/n). After two periods, P(1+r/n)2P \cdot (1+r/n)^2. After ntnt total periods, P(1+r/n)ntP \cdot (1+r/n)^{nt}. This is the constant-ratio story in action: each period multiplies by (1+r/n)(1 + r/n).

Half-life: the canonical decay story. A radioactive substance with half-life HH has amount: A(t)=A0(12)t/H=A02t/H.A(t) = A_0 \cdot \left(\frac{1}{2}\right)^{t/H} = A_0 \cdot 2^{-t/H}.

After every HH time units, the amount halves — constant ratio 1/21/2 per HH years.

Exponentials vs polynomials: the massacre

Here is a numerical demonstration. I want you to watch carefully because the Federation of Boring Textbook Authors either skips this or buries it in a footnote, and that is a crime. Let f(x)=x100f(x) = x^{100} (a ferocious polynomial) and g(x)=2xg(x) = 2^x (a humble exponential with the smallest non-trivial integer base).

xxx100x^{100}2x2^x
10101010010^{100}10241024
100010001030010^{300}21000103012^{1000} \approx 10^{301}
10000100001040010^{400}2100001030102^{10000} \approx 10^{3010}

The polynomial is winning massively at x=10x = 10. By x=1000x = 1000 it's a tie. By x=10000x = 10000, the exponential has lapped the polynomial so many times it's embarrassing. The precise statement: for any polynomial p(x)p(x) and any base b>1b > 1,

limxp(x)bx=0.\lim_{x \to \infty} \frac{p(x)}{b^x} = 0.

This is a calculus result (L'Hopital repeated application), but the numerical table makes it undeniable. Exponentials eventually crush every polynomial. No exceptions, no negotiations. File this away; it will matter in computer science (time complexity), finance, biology, and physics, and anyone who tells you otherwise is selling you bullshit.

The number ee: the natural base

There is one base that is mathematically natural in a way no other base is. Consider the question: for what base bb does the exponential function bxb^x have the property that its *rate of change at x=0x = 0 equals exactly 11?

Answer: b=eb = e, where: e=limn ⁣(1+1n)n2.71828e = \lim_{n \to \infty}\!\left(1 + \frac{1}{n}\right)^n \approx 2.71828\ldots

This is irrational (and in fact transcendental — a fact that requires significant machinery to prove). You can think of ee as the continuous-compounding limit: if you compound interest continuously instead of finitely many times per year, the 11 dollar balance grows to ere^r dollars after one year at rate rr.

The exact reason ee is natural — that ddxex=ex\frac{d}{dx}e^x = e^x (the derivative of exe^x is itself) — requires calculus. But the number ee and the function ex=exp(x)e^x = \exp(x) appear now because you need them for logarithms in the next lesson. You will see ee in every stratum from here to the spectral theorem, lurking in every corner of mathematics like a brilliant and slightly unhinged colleague who shows up uninvited to every party and is always right. Treat it with respect.

🔬 SPECIMENS (worked examples)

Worked example 1 — identifying exponential from a table

A table of values is given: xf(x)061182543162\begin{array}{c|c} x & f(x) \\ \hline 0 & 6 \\ 1 & 18 \\ 2 & 54 \\ 3 & 162 \end{array} Show this is exponential and write its formula.

Test for constant ratio. Compute consecutive output ratios: f(1)f(0)=186=3,f(2)f(1)=5418=3,f(3)f(2)=16254=3.\frac{f(1)}{f(0)} = \frac{18}{6} = 3, \quad \frac{f(2)}{f(1)} = \frac{54}{18} = 3, \quad \frac{f(3)}{f(2)} = \frac{162}{54} = 3.

Constant ratio b=3b = 3 at every step. This is exponential.

Formula. General form: f(x)=abxf(x) = a \cdot b^x. We know b=3b = 3 and f(0)=a30=a=6f(0) = a \cdot 3^0 = a = 6. So: f(x)=63x.f(x) = 6 \cdot 3^x.

Check: f(2)=69=54f(2) = 6 \cdot 9 = 54. Correct.

Worked example 2 — compound interest calculation

You deposit \1000atanannualinterestrateofat an annual interest rate of6%$, compounded monthly. How much do you have after 5 years?

Formula: A=P ⁣(1+rn)ntA = P\!\left(1 + \frac{r}{n}\right)^{nt}.

Here P=1000P = 1000, r=0.06r = 0.06, n=12n = 12 (monthly), t=5t = 5.

A=1000 ⁣(1+0.0612)125=1000(1.005)60.A = 1000\!\left(1 + \frac{0.06}{12}\right)^{12 \cdot 5} = 1000 \cdot (1.005)^{60}.

Compute (1.005)60(1.005)^{60}. Using ln(1.005)0.004988\ln(1.005) \approx 0.004988, we get 60ln(1.005)0.299360 \ln(1.005) \approx 0.2993, so (1.005)60e0.29931.3489(1.005)^{60} \approx e^{0.2993} \approx 1.3489.

A10001.3489=$1,348.90.A \approx 1000 \cdot 1.3489 = \$1{,}348.90.

Note: this is more than 1000 \cdot (1 + 0.06 \cdot 5) = \1300$ (simple interest), because compounding adds interest-on-interest.

Worked example 3 — the exponential always wins, proved by brutal honest arithmetic

Compare the growth of p(x)=x3p(x) = x^3 and g(x)=1.5xg(x) = 1.5^x for large xx. Find (approximately) where gg first overtakes pp.

We need to find where 1.5x>x31.5^x > x^3.

At x=10x = 10: p(10)=1000p(10) = 1000, g(10)=1.51057.7g(10) = 1.5^{10} \approx 57.7. Polynomial winning.

At x=20x = 20: p(20)=8000p(20) = 8000, g(20)=1.5203325g(20) = 1.5^{20} \approx 3325. Still polynomial winning.

At x=30x = 30: p(30)=27000p(30) = 27000, g(30)=1.530191751g(30) = 1.5^{30} \approx 191751. Exponential winning by a factor of 7.

The crossover is between x=20x = 20 and x=30x = 30. Let's try x=25x = 25: p(25)=15625p(25) = 15625, g(25)=1.52525251g(25) = 1.5^{25} \approx 25251. The exponential wins at x=25x = 25.

So even with the slow-growing base 1.51.5, the exponential overtakes x3x^3 around x24x \approx 24. After that, the polynomial has no chance: gg pulls further and further ahead, forever. This is the meaning of "exponentials eventually dominate every polynomial."

☠ KNOWN HAZARDS

  • Confusing 2x2^x with x2x^2. These are completely different functions and confusing them is the kind of mistake that keeps me up at night. x2x^2 is a polynomial (exponent fixed, base variable). 2x2^x is exponential (base fixed, exponent variable). The growth rates differ catastrophically for large xx. Do not mix them up.

  • Thinking base b=1b = 1 is an exponential. f(x)=1x=1f(x) = 1^x = 1 is a constant function, not exponential. The interesting case requires b1b \ne 1. This should be obvious, but here we are.

  • Forgetting the horizontal asymptote. The function f(x)=2xf(x) = 2^x never actually equals zero. It approaches zero asymptotically. "The xx-axis is an asymptote" and "the function crosses the xx-axis" are mutually exclusive. Approach is not contact.

  • Misapplying compound interest. The exponent is ntnt (number of periods), not tt alone. Annual compounding (n=1n = 1): (1+r)t(1 + r)^t. Monthly (n=12n = 12): (1+r/12)12t(1 + r/12)^{12t}. Don't mix up tt (years) and ntnt (total periods). This kind of error costs people real money in the real world, so get it right.

TL;DR

  • An exponential f(x)=abxf(x) = a \cdot b^x has a constant multiplicative ratio bb between successive outputs — every step multiplies by bb.

  • Graph: always positive, through (0,a)(0, a), horizontal asymptote y=0y = 0; growth if b>1b > 1, decay if 0<b<10 < b < 1.

  • Compound interest A=P(1+r/n)ntA = P(1 + r/n)^{nt} and half-life A=A0(1/2)t/HA = A_0 (1/2)^{t/H} are the canonical applications.

  • For any polynomial p(x)p(x) and base b>1b > 1: eventually bxb^x dominates p(x)p(x). Exponential growth beats polynomial growth.

  • e2.718e \approx 2.718 is the natural base, arising from continuous compounding; exe^x is the most important exponential function.

unlocks